Where does the energy go in fs-relaxation?

- Vibronic excitation to a steep slope leaves only little energy in this active mode.
- The missing energy disappears as fast as the wave packet travels down (≈50 fs).
- Larger molecules have longer time constants. Equipartition is not reached in ≈5 ps.
- Underway several coordinates begin to exhibit steep slopes and share the energy.
- For example, in ethene besides the twist also 7-11 more coordinates contribute.

In published gas-phase femtosecond spectroscopy of dimethylaminobenzonitrile, unsaturated metal carbonyls M(CO)5 and ethylene, coherent oscillations were observed that were assigned to vibrational states near potential minima, although the preceding relaxation released much more energy than found in the vibrations. The missing kinetic energy disappeared from the active mode in times around 50 fs, but then the oscillations were observed over several picoseconds. This as well as molecular size effects are difficult to explain by models based on perturbation theory and statistics. A closer inspection of the examples reveals, however, more coordinates with steep slopes. They can share the energy. Other modes do not take part, and there is no equipartition until several picoseconds. It is conceivable that also in high-level calculations redistribution of energy was a bottleneck. It is suggested to check them also for oscillations and for isotope effects, which give indications where the kinetic energy is localized.